The Inner Workings of LLMs: A Deep Dive into Language Model Architecture